You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently in OpenMC, both the MGMC mode and random ray both require the user to input multigroup cross sections. This is a generally good policy given the approximations inherent with MGXS data.
However, as OpenMC is also capable of producing MGXS from a continuous energy mode, it could be nice to set up an automated workflow internal to OpenMC that would would perform a coarse-grained CE transport stage at the beginning to generate approximated MGXS data before beginning a multigroup solve based on that data. The upside is that the user could define their materials using standard isotopic content lists in the same way as with continuous energy MC, and wouldn't need to manually set up two separate models (one to generate MGXS with CE, and then another to read in the MGXS and use that).
The goal here is to reduce the barrier to entry for the multigroup solvers. Currently, if a user is just wondering if MGMC or random ray is going to be faster than MC for their use case, they first have to jump through a bunch of hoops to come up with MGXS data. If this pipeline was automated, then hopefully answering this question could be as easy as just setting 2-3 additional parameters in their settings config of their usual CE MC model. If they like the performance values they are seeing, then they can crank up the fidelity of the preceding CE MC solve to generate better MGXS, or perhaps change a setting to increase the spatial resolution of the cross sections from material-wise --> cell-wise --> material filled cell instance-wise, etc.
Ideally, the MGXS data generated via this pipeline could be stored to file for later re-use, but then a flag in the settings can be set such that the user doesn't have to go in and redefine all their materials again to use the pre-generated data. Basically, there needs to be a way for the user to easily say "I have defined my materials with isotopic content lists, please generate MGXS data" vs. "I have defined my materials with isotopic content lists, but you've already generate MGXS for them, so read those from file". Basically, we don't want the user to have to go in and redefine all their materials as MGXS materials instead of CE content list materials each time they want to switch back and forth between CE mode vs. MGXS mode.
The text was updated successfully, but these errors were encountered:
Currently in OpenMC, both the MGMC mode and random ray both require the user to input multigroup cross sections. This is a generally good policy given the approximations inherent with MGXS data.
However, as OpenMC is also capable of producing MGXS from a continuous energy mode, it could be nice to set up an automated workflow internal to OpenMC that would would perform a coarse-grained CE transport stage at the beginning to generate approximated MGXS data before beginning a multigroup solve based on that data. The upside is that the user could define their materials using standard isotopic content lists in the same way as with continuous energy MC, and wouldn't need to manually set up two separate models (one to generate MGXS with CE, and then another to read in the MGXS and use that).
The goal here is to reduce the barrier to entry for the multigroup solvers. Currently, if a user is just wondering if MGMC or random ray is going to be faster than MC for their use case, they first have to jump through a bunch of hoops to come up with MGXS data. If this pipeline was automated, then hopefully answering this question could be as easy as just setting 2-3 additional parameters in their settings config of their usual CE MC model. If they like the performance values they are seeing, then they can crank up the fidelity of the preceding CE MC solve to generate better MGXS, or perhaps change a setting to increase the spatial resolution of the cross sections from material-wise --> cell-wise --> material filled cell instance-wise, etc.
Ideally, the MGXS data generated via this pipeline could be stored to file for later re-use, but then a flag in the settings can be set such that the user doesn't have to go in and redefine all their materials again to use the pre-generated data. Basically, there needs to be a way for the user to easily say "I have defined my materials with isotopic content lists, please generate MGXS data" vs. "I have defined my materials with isotopic content lists, but you've already generate MGXS for them, so read those from file". Basically, we don't want the user to have to go in and redefine all their materials as MGXS materials instead of CE content list materials each time they want to switch back and forth between CE mode vs. MGXS mode.
The text was updated successfully, but these errors were encountered: